Using hybrid information of colour image analysis and SWIR-spectrum for high-precision analysis of construction and demolition waste

This paper discusses the accuracy improvement of automatic analysis of construction and demolition waste (CDW) by using the combination of image analysis and spectral information. This means using the combination of methods of image processing, methods of spectral analysis and methods of supervised learning. The classification performances in colour images and also in SWIR-spectrums showed, that we have to use a combination of these two components in a combined feature vector to improve the accuracy of analysis. Investigations on hybrid information from colour images and SWIR-spectrums were done and compared with the separate usage of these information sources.